fix(db): recover failed credential group index on migration retry

[TheodoreSpeaks]

Summary

• Recover the invalid workspace credential-group index left by a failed concurrent build when migration 0326 is retried.
• Preserve healthy indexes and existing data. Remaining duplicate groups still fail uniqueness validation.
• Fix a flaky connector test to compare exact IDs instead of rejecting incidental substring matches in generated IDs.

Type of Change

• Bug fix

Testing

Six PostgreSQL migration tests passed, covering duplicate failures, recovery after removing duplicates, interrupted recovery, and healthy-index preservation. All 58 connector orchestration tests passed. Database type checking, lint, all 46 audits, docs-manifest validation, and migration safety checks passed.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated
docs	Skipped		Sep 8, 2026 3:19pm UTC

[greptile-apps]

Greptile Summary

This PR makes migration 0326 recover automatically from an invalid workspace credential-group index left by a failed concurrent build.

• Renames only the expected invalid index before dropping it concurrently.
• Refuses to remove a valid or otherwise unexpected index occupying the recovery name.
• Adds PostgreSQL coverage for duplicate-data retries, interrupted recovery, and healthy-index preservation.
• Corrects a connector test to compare identifiers exactly.

Confidence Score: 5/5

The PR appears safe to merge; no outstanding correctness or repository-rule issue was identified.

The migration narrowly recovers the expected invalid index, rejects unexpected recovery-name occupants, retains concurrent index operations, and has focused replay coverage for its important failure states. The follow-up connector assertion remains safe and tests the intended identifier contract more precisely.

Important Files Changed

Filename	Overview
packages/db/migrations/0326_enterprise_organization_search.sql	Adds guarded, replay-safe recovery for the invalid workspace uniqueness index while preserving healthy indexes and existing rows.
packages/db/organization-search-migration.postgres.test.ts	Expands PostgreSQL migration tests across duplicate failure, repaired retry, interrupted recovery, and healthy-index replay scenarios.
apps/sim/lib/knowledge/orchestration/connectors.test.ts	Replaces substring matching with an exact connector-identifier inequality assertion.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Retry migration 0326] --> B{Recovery-named index exists?}
    B -- Yes --> C{Invalid and owned by credential_group?}
    C -- No --> D[Abort without dropping index]
    C -- Yes --> E[Continue recovery]
    B -- No --> E
    E --> F{Original workspace index is invalid?}
    F -- Yes --> G[Rename to recovery name]
    F -- No --> H[Preserve original index]
    G --> I[Drop recovery index concurrently]
    H --> I
    I --> J[Create workspace unique index concurrently if absent]
    J --> K{Duplicate workspace groups remain?}
    K -- Yes --> L[Uniqueness build fails; data remains]
    K -- No --> M[Valid unique index restored]

Loading

_{Reviews (2): Last reviewed commit: "fix(test): compare connector identifiers..." | Re-trigger Greptile}

[TheodoreSpeaks]

@greptile

[TheodoreSpeaks]

@cubic-dev-ai review this PR

[cubic-dev-ai]

@cubic-dev-ai review this PR

@TheodoreSpeaks I have started the AI code review. It will take a few minutes to complete.